It's a classic scenario in a recruitment agency. A mission comes in for a rare profile. You open LinkedIn, spend three days headhunting, and finally approach a promising candidate. Then a colleague tells you: "Oh yes, I met her two years ago."
She was in your database. You just didn't find her.
That's the paradox of modern sourcing: most agencies and executive search firms start from scratch on every mission, even though they're sitting on thousands of already-qualified profiles. Good sourcing software isn't just for finding new profiles outside your walls. Its first job is to help you find the ones you already have, then go get the rest.
What to remember
- Sourcing software identifies profiles both in your internal talent pool and outside it. Don't confuse it with an ATS, which manages inbound applications: sourcing is proactive, an ATS is reactive.
- The real goldmine is internal. 68% of companies keep applications on file, but barely a third re-review them for a new hire. Fewer than half of recruiters actually make use of their talent pool.
- Keyword search misses good profiles: it requires guessing the candidate's exact vocabulary. Semantic search understands intent and surfaces profiles whose CV doesn't use your words.
- The 5 selection criteria: natural-language search, internal talent pool queried first, automatic data freshness, native ATS and CRM integration, GDPR compliance.
- With Marvin, Scout searches your database first (CVs, notes, history), then enriches from LinkedIn. The Talent Refresh agent keeps the data up to date, and the profile found lands directly in the Desk pipeline, with no re-entry.
What is sourcing software?
Sourcing software is a tool that identifies and finds candidate profiles, in your own talent pool as well as outside it (LinkedIn, CV databases, the web), based on a hiring need. Where an ATS manages the applications that come to you, candidate sourcing software actively looks for profiles that haven't come forward.
The distinction is often misunderstood.
- The ATS is reactive: it centralizes incoming applications, manages the process, tracks interviews. It works on an inbound flow.
- Sourcing software is proactive: it queries a database (internal or external) to surface profiles matching a need. It works on an existing stock.
The two are complementary, and ideally they share the same data as your ATS and your CRM, to avoid re-entering information between tools.
Good recruitment sourcing software therefore covers two distinct areas:
- Internal sourcing: finding candidates in your talent pool you've already met, already qualified, sometimes already presented to another client
- External sourcing: identifying new profiles on LinkedIn, CV databases and the web, then bringing them into your database
In practice, most tools on the market focus on the second and neglect the first. That's precisely where the most value is lost, and it's the subject of this article.
The real problem: your talent pool is asleep
Before looking for a tool, you need to face the problem head-on. Most agencies accumulate profiles for years without ever reactivating them.
The numbers speak for themselves (French market data):
- 68% of companies keep the applications they receive, but barely a third of them re-review them for a new hire.
- Fewer than half of recruiters actually make use of their talent pool, even though 82% believe an active talent pool cuts time-to-hire by more than 40%.
In other words: everyone knows the talent pool is an asset, almost nobody uses it. Three reasons explain this gap.
1. The data ages fast
A profile entered two years ago is no longer current. The candidate has changed jobs, moved, earned a certification, or become available again without your knowledge. A talent pool that isn't refreshed becomes an archive, not a working tool.
2. Keyword search misses good profiles
This is the most underestimated cause. You're looking for a "national account manager," the CV says "head of strategic development." The profile exists, it matches, but your query doesn't surface it. On rare profiles, vocabulary varies enormously from one CV to another, and that's exactly where classic search hits its limit.
3. Nobody has time to dig
A talent pool of 5,000 profiles with no effective search tool is a talent pool of 0 usable profiles. Faced with an urgent mission, the reflex is to go to LinkedIn rather than spend an hour exploring your own database. The cost of internal search is perceived as higher than external headhunting, which is an admission of tooling failure.
For an executive search firm, this waste is even more pronounced. Every profile met represents relationship capital built on hours of conversation. Failing to find them again when a mission finally matches is losing twice: the time already invested, and the opportunity at hand.
Boolean search or semantic search: the difference that changes everything
This is the most technical point in this article, and also the most decisive. The two approaches aren't looking for the same thing.
What keyword search does
Boolean search, the kind offered by most ATSs and LinkedIn Recruiter, searches for strings of characters. You write a query with operators (AND, OR, NOT, quotation marks, parentheses) and the tool returns the profiles that contain exactly those terms.
It works well when you know precisely which words appear on the CV. Its limits show up fast:
- It requires guessing the candidate's vocabulary. The same job title is written ten different ways depending on the industry, the company and the seniority of the profile.
- It doesn't understand synonyms or context. "P&L management" and "budget responsibility" describe the same thing, but not to a Boolean engine.
- It demands a skill in itself. Writing a good Boolean query is the know-how of an experienced sourcer, not second nature for a junior consultant.
- It doesn't rank. You get a list of profiles that contain the words, not a list of profiles relevant to your mission.
What semantic search does
Semantic search understands the intent behind the query. You describe your need in natural language, the way you'd phrase it to a colleague, and the engine analyzes the meaning of the candidate's career path rather than the presence of exact words.
In practice, you can write: "a CFO who led a fundraising round at a growing SaaS company, open to relocating to Lyon." The engine will return profiles whose CV may not contain any of those terms, but whose background genuinely matches.
What that changes day to day:
- You describe the need, you don't guess the vocabulary
- Profiles are ranked by relevance, not listed by literal match
- Any consultant can search effectively, without Boolean training
- Older profiles resurface: a candidate met three years ago comes back up if their background matches, even if nobody remembers them
This is also known as AI sourcing: artificial intelligence doesn't replace the recruiter, it gives them a better map of their own database.
Why it's decisive for an executive search firm
In executive search, the gap between the two methods becomes a competitive edge. Executive profiles have unusual career paths, described in vocabulary specific to each industry. Two CFOs who did exactly the same job will present lexically very different CVs.
The first firm to present a relevant shortlist often wins the mission. When internal search takes 30 seconds instead of two days of LinkedIn headhunting, that's no longer just convenience: it's commercial speed.
The 5 criteria of good sourcing software
Here's the checklist to apply against any solution on the market.
| Criterion | What to check | Why it's decisive |
|---|---|---|
| Semantic search | Can you describe a need in natural language, without Boolean operators? | This is what surfaces profiles whose CV doesn't use your words |
| Internal talent pool first | Does the tool search your database before going outside? | A candidate you've already met costs less to re-engage than a stranger to convince |
| Data freshness | Do profiles update automatically (role, availability)? | An unrefreshed talent pool becomes an archive within 2 to 3 years |
| ATS and CRM integration | Does the profile found land directly in your pipeline? | Otherwise you're adding one more tool, and one more re-entry |
| GDPR compliance | Where is the data hosted? What legal basis, what retention period? | A candidate talent pool is personal data processing, with a duty to inform |
Two criteria deserve an extra word.
Integration is the classic trap. Many sourcing tools are excellent in isolation, but run alongside your ATS. Result: you find a profile, export it, re-import it, retype it. Every break between two tools is a chance to lose data, and often the candidate along with it.
Compliance isn't an administrative detail. Building a talent pool means informing candidates, defining a retention period and enabling them to exercise their rights. For a company operating under GDPR, data hosted in Europe and compliance by design remove the need to worry about it after the fact.
Sourcing with Marvin: the talent pool first, LinkedIn second
Our approach lies in the order of operations. Most sourcing tools send you to search outside first. We start by looking at what you already have.
A natural-language search on your own database
With Scout, our sourcing app, you describe the profile you're looking for the way you'd describe it to a colleague. No operators, no syntax to master. Semantic search analyzes your entire talent pool: CVs, interview notes, conversation history, imported LinkedIn profiles.
In practice, that means:
- A candidate met three years ago resurfaces if their background matches today's mission
- Your colleagues' notes are searchable, not just CVs
- Any consultant can search effectively, including a junior who started last week
That's the difference between a CV database and a talent pool you can actually use.
External enrichment, once the talent pool is exhausted
When your database isn't enough, Scout searches outside: live profile search on LinkedIn and the web, with contact enrichment. New profiles join your talent pool, where they'll stay available for future missions.
Every hunt therefore adds to your asset instead of evaporating once the mission closes.
A talent pool that maintains itself
That's the role of our Talent Refresh agent: it continuously enriches the talent pool and detects changes in status, notably profiles that switch to "Open to Work." You don't start over from a frozen database on every mission, you work on data that's alive.
The profile found goes straight into the pipeline
Because Scout shares the same data as Desk, our ATS and CRM, the candidate you identify lands in your process, not in an Excel file. No export, no re-import, no re-entry. That's what sets an integrated suite apart from a stack of tools.
At a glance, here's what changes:
| Classic sourcing | Sourcing with Marvin | |
|---|---|---|
| Starting point | LinkedIn, blank slate on every mission | Your internal talent pool first |
| Search method | Boolean query to build | Description of the need in natural language |
| Older profiles | Forgotten, invisible | Resurface if they match |
| Scope analyzed | CV only | CV, interview notes, history, LinkedIn |
| Data freshness | Frozen at entry date | Maintained by Talent Refresh |
| After the search | Export, re-import, re-entry | The profile is already in the pipeline |
The gain isn't just convenience. Our clients replace an average of 5 to 7 tools with a single platform and free up around 7 hours per week per consultant on low-value tasks. For a 10-consultant agency, that's the equivalent of two production days handed back every week.
And for those who already work with AI daily, Marvin can also be used directly from Claude or ChatGPT via its MCP server: you ask for a shortlist in your usual chat, and it's built from your talent pool.
Your talent pool isn't an archive, it's your asset
Back to the candidate from the start of this article, the one asleep in your database while you searched for her on LinkedIn. This isn't a rigor or organization problem. It's a tooling problem.
As long as finding an internal profile costs more effort than looking for a new one outside, your consultants will always go outside. That's rational on their part, and it's what drains your talent pool of its value year after year.
Good sourcing software reverses that logic. When internal search happens in natural language, in seconds, on data kept up to date, the reflex changes on its own: you look at what you already have, first. Profiles met two or three years ago become presentable candidates again, and every new hunt adds to a growing asset instead of evaporating.
Whether you run a recruitment agency or an executive search firm, the difference is measured where it counts: how fast you produce a relevant shortlist, and therefore how many missions you win.
Want to see what's actually in your talent pool? Book a demo: we'll show you, in real conditions, how a natural-language search surfaces profiles from your own database that you'd forgotten about, and how migrating your data from your current tool works.
Frequently asked questions about sourcing software
What's the difference between an ATS and sourcing software?
The ATS is reactive, sourcing software is proactive. The ATS centralizes the applications that come to you and manages process tracking: interviews, statuses, decisions. Sourcing software actively looks for profiles that haven't come forward, in your internal talent pool as well as outside it. The two are complementary, and ideally they share the same database: otherwise you export and re-import profiles between two tools, with the data loss that implies.
What is AI sourcing?
AI sourcing refers to the use of artificial intelligence to identify candidates, mainly through semantic search. Where a Boolean search looks for exact words, AI analyzes the meaning of a career path: it understands that a "head of strategic development" can match a search for "national account manager." It also ranks results by relevance rather than literal match. AI doesn't replace the recruiter's judgment, it gives them better visibility into available profiles, starting with those already in their own database.
How do you source candidates on LinkedIn effectively?
LinkedIn remains essential, but the common mistake is going there first. The right order is to exhaust your internal talent pool first, where profiles are already qualified and often already met, before hunting outside. Then, on LinkedIn, effectiveness comes down to three things: a precise query, a personalized outreach message explaining why this particular candidate, and above all systematically bringing identified profiles back into your database. A sourced profile that stays on LinkedIn is lost for the next mission.
Is sourcing software useful for an executive search firm?
That's actually where it delivers the most value. In executive search, executive career paths are unusual and described in vocabulary specific to each industry: two CFOs who held the same responsibilities will have lexically very different CVs. Keyword search is particularly ineffective here. On top of that comes relationship capital: every profile met represents hours of conversation. Failing to find them again when a mission finally matches means losing both the initial investment and the opportunity.
Is AI sourcing GDPR compliant?
It can be, provided it meets the same obligations as any candidate data processing: informing individuals, defining a legal basis and a retention period, enabling the exercise of access and deletion rights. The points to check with a vendor are data hosting, compliance by design and processing traceability. A candidate talent pool is personal data processing, not a simple address book: it's a selection criterion in its own right, not a formality.
